There is not a public record to be examined and ruminated on, on the contrary it's a situation where you can go and see "oh 151,400 subscribers to fatpeoplehate, this is normal behavior", or (more likely) people who never view comments will just start to see jokes about killing the obese on their front page feed.
There's a difference between 'saving people from unpleasant speech' and providing an echo chamber to hate speech along side the company's incredibly popular other content.
I don't see the connection to antivaxxer logic, but I'll give you the benefit of the doubt there?
Just to be clear I'm not advocating that we ban all 'unpleasant speech' from the internet, but I do think siloing it off so you really have to be looking for it works, and that there's significant evidence to back that up.
In reddit's case in particular, many of the banned subreddits were using smart tactics like brigading and vote manipulation to amplify their numbers, and push the 'this is normal' message even further, although that's somewhat orthogonal to what we're arguing about.
And if a community deletes or down-votes dissent: well, that happens but it also becomes a matter of public record and can also inform. Case in point: the moderation of /r/politics.
Nothing is served by "saving" Joe User from being exposed to unpleasant speech in an open forum, due to the fear that he might be adversely affected or even damaged by the experience. This is anti-vaxxer logic, on a meme level.